| 02:26 | <dglazkov> | ping Hixie |
| 02:26 | <dglazkov> | with a cherry on top |
| 02:28 | dglazkov | yodels in hopes of attracting attention |
| 02:31 | <dglazkov> | alas |
| 08:09 | <gsnedders> | There are no bugs, only features. |
| 09:37 | Philip` | notices that http://www.robbyslaughter.com/blog/?2008-03-21 , saying "You might have noticed that not all websites work perfectly", does not work in Opera since it renders as white text on a white background |
| 09:43 | <jruderman_> | i wonder why that happens |
| 09:44 | <Philip`> | I'd guess it's to do with style inheritance into tables in quirks mode, which no browsers are particularly consistent on |
| 09:48 | <hsivonen> | Unicode and Math really messed up with φ. Not good. |
| 09:49 | <othermaciej> | hsivonen: how so? |
| 09:49 | <hsivonen> | Unicode swapped the glyphs for two code points around. |
| 09:50 | <hsivonen> | then Math swapped the entities around to keep the entity-glyph mapping |
| 09:50 | <hsivonen> | now φ means different things in Gecko depending on doctype |
| 09:51 | <othermaciej> | oy vay |
| 10:26 | <hsivonen> | FYI: http://groups.google.com/group/mozilla.dev.tech.mathml/browse_thread/thread/e7f7efbb5e161348/9fde74f46fb0b5d2#9fde74f46fb0b5d2 |
| 10:28 | <othermaciej> | not sure if preventing grief for Safari is an argument that will fly :-) |
| 10:28 | <othermaciej> | but that does sound like a good idea |
| 10:31 | <BenMillard> | I read the excessive DTD traffic article a few days ago...found it funny in a tragic sort of way. |
| 10:57 | <hsivonen> | othermaciej: I'm hoping it works in the "do unto others" sense and with the Mission of the Foundation |
| 12:26 | <hsivonen> | looks like the TAG issue list has even older items than the WHATWG issue list... http://www.w3.org/2001/tag/issues.html?view=normal&closed=1&type=1 |
| 12:41 | <MikeSmith> | hsivonen - do you keep up with the httpbis list? |
| 12:42 | gsnedders | sighs at the very mention of that list |
| 12:42 | <hsivonen> | MikeSmith: I don't. |
| 12:43 | <hsivonen> | MikeSmith: is there something going on there that I should be aware of? |
| 12:43 | <MikeSmith> | not that I know of. was just asking because I've got a bunch of unread messages from that and trying to figure out what threads to read |
| 12:44 | <hsivonen> | I see IDNAbis in the subjects. That can't be good. |
| 12:44 | <MikeSmith> | heh |
| 12:44 | <MikeSmith> | yep |
| 12:44 | <gsnedders> | hsivonen: the recent discussion is all idealistic as if compat. with current non-conforming impls is irrelevant |
| 12:47 | <hsivonen> | eek. Are they really contemplating allowing non-ASCII on the HTTP "GET" line? |
| 12:48 | <gsnedders> | I dunno. I gave up reading it. |
| 12:48 | <gsnedders> | :) |
| 12:49 | <hsivonen> | It's one thing to insist on a multitude of human scripts in user-readable prose and to insist on non-ASCII in computer-to-computer messages that are text-based for programmer convenience |
| 12:50 | <hsivonen> | s/and/and another/ |
| 12:51 | hsivonen | thinks people who advocate non-ASCII in programming language variable/method names or XML element/attribute names are misguided |
| 13:35 | <hsivonen> | hmm. does SVG really have no IDREF or IDREFS attributes? |
| 13:41 | <hsivonen> | do UAs actually implement SVG <use> as a DOM tree clone or as an anonymous layout object tree? |
| 13:41 | <hsivonen> | or something else? |
| 13:42 | <hsivonen> | "Property inheritance, however, works as if the referenced element had been textually included as a deeply cloned |
| 13:42 | <hsivonen> | child of the 'use' element." |
| 13:42 | <hsivonen> | that sounds really bad |
| 13:45 | <hsivonen> | hmm. looks like the SVG WG has created a need for a hashed ID reference in their spec, as well |
| 13:46 | <hsivonen> | "● local URI references, where the URI reference does not contain an <absoluteURI> or <relativeURI> and |
| 13:46 | <hsivonen> | thus only contains a fragment identifier (i.e., #<elementID> or #xpointer(id<elementID>)) |
| 13:46 | <hsivonen> | " |
| 13:46 | <hsivonen> | is the xpointer alternative interoperable? |
| 13:47 | <hsivonen> | also, in compound documents, can are "local" references local to the SVG subtree or to the entire document? |
| 14:00 | <hsivonen> | are cross-document <use> and <tref> supposed to work? |
| 14:02 | <Philip`> | hsivonen: http://krijnhoetmer.nl/irc-logs/whatwg/20080311#l-655 is relevant about implementation of SVG <use> |
| 14:04 | <hsivonen> | Philip`: yeah. I'm trying to figure out if I should try to validate it as a mere syntactic IRI or as a hashed ID reference whose referent is actually checked for sanity |
| 14:05 | <hsivonen> | I do agree with othermaciej's assessment |
| 17:23 | <met_> | Just found thant input.click() in some input types elements doesn't call the onclick handler in Firefox. Strange. |
| 17:23 | <met_> | testcase http://a.met.cz/oreilly/flanagan/ |
| 17:24 | <met_> | it ignores type text, password, file, hidden - all textlike input elements |
| 18:37 | <a-ja> | Hixie: ping (when you're done chatting with Bernd....no hurry or anything) |
| 18:45 | <Hixie> | yo |
| 18:46 | <a-ja> | was working on html5'izing a site other day, & noticed i couldn't get valid dir listing pages on an apache 1.3 server...even using custom header and readme files. prob is that they get a pre with a hr inside. |
| 18:47 | <a-ja> | i presume that was valid html 3.2 |
| 18:48 | <a-ja> | can get em to work with apache 2.something options. worth a special case in 5? |
| 18:49 | <Hixie> | nah |
| 18:49 | <Hixie> | those pages don't have an html5 doctype anyway |
| 18:49 | a-ja | has to get a _real_ hoster :) |
| 18:50 | <a-ja> | not by default they don't anyway |
| 18:51 | <a-ja> | tend to agree......real edge case |
| 18:53 | <a-ja> | 1.3's kind of a dinosaur already (but plenty of em still around, i'd imagine) |
| 18:57 | <a-ja> | on unrelated matter...filed a html5-related moz 1.9 bug the other day you might wanna cc yourself on...which cropped up with the recent parser change |
| 18:58 | <a-ja> | bug 423721 legend child of figure or unknown element should not generate fieldset |
| 18:59 | <a-ja> | feel free to steal /modify the testcase there if you want |
| 20:22 | <Philip`> | a-ja: About "plenty of em still around": Counting 130K pages from dmoz.org, I see 33K Apache 1.3, 16K of 2.0, 9K of 2.2, 23K unlabelled Apache |
| 20:22 | <Philip`> | so I'd agree there's plenty |
| 20:23 | <Philip`> | (Also I see 22K IIS 6.0, 7K 5.0, negligible other versions) |
| 20:41 | <hsivonen> | it irks me that the SVG 1.1 and 1.2 Tiny specs don't tell me if they expect <use> and <tref> to work across documents. |
| 20:42 | <hsivonen> | I guess I have to write a test case and find out. |
| 20:52 | <hsivonen> | findings so far: Cross-doc doesn't work in Firefox, but the impl does URI resolution and isn't just like the hashed ID ref |
| 20:53 | <hsivonen> | SVG still quits Safari on me. Odd |
| 20:54 | <othermaciej> | hsivonen: do you have a specific SVG that does it? |
| 20:54 | <hsivonen> | othermaciej: yes, I have |
| 20:55 | <hsivonen> | Opera appears to implement real cross-doc <use> |
| 20:55 | <othermaciej> | hsivonen: can you file a bug, or mail it to me? |
| 20:55 | <hsivonen> | othermaciej: I've already filed a bug |
| 20:55 | <othermaciej> | hsivonen: I wouldn't be surprised if your <use> impl has bugs |
| 20:55 | <othermaciej> | I freaking hate <use> |
| 20:55 | <othermaciej> | hsivonen: thanks! |
| 20:56 | <hsivonen> | othermaciej: unfortunately, the information I've provided isn't very useful |
| 20:56 | <othermaciej> | hsivonen: if you included the file that crashes, that's useful enough |
| 20:57 | <hsivonen> | oops. I've missed a new comment in the bug |
| 20:59 | hsivonen | tries to get a backtrace now |
| 21:03 | <hsivonen> | whoa. I have the Adobe Plug-in |
| 21:03 | <Hixie> | http://scripts.indisguise.org/2008/03/22/xhtml-versus-html/ |
| 21:05 | <othermaciej> | hsivonen: oh - we'll use ASVG when it's installed |
| 21:05 | <othermaciej> | I think maybe we should stop preferring it without explicit opt-in |
| 21:05 | <hsivonen> | commented: http://bugs.webkit.org/show_bug.cgi?id=17073 |
| 21:06 | <hsivonen> | sorry about blaming WebKit. I had no idea I had Adobe residue on the system |
| 21:07 | <hsivonen> | looks like WebKit implement <use xlink:href> as a hashed ID reference |
| 21:07 | <hsivonen> | Firefox does URI-resolution and if it points to a fragment in the same doc, uses the fragment |
| 21:08 | <hsivonen> | Opera actually allows references to different HTTP resources |
| 21:08 | <hsivonen> | yay for interop |
| 21:08 | <hsivonen> | so should Validator.nu allow a full IRI, a full URI or require a hashed ID ref? |
| 21:09 | <hsivonen> | actually, I should rerun the Opera and Firefox tests in case either loaded the Adobe plug-in |
| 21:11 | <hsivonen> | those results hold |
| 21:12 | <a-ja> | Philip`: tks for #'s (was AFK) |
| 21:14 | <hsivonen> | Hixie: I'd love to see some numbers proving an XML parser marginally faster especially in the case that it is reading a DTD (even an abridged local one) |
| 21:16 | <othermaciej> | I am pretty sure that in WebKit the XML parser is slower than the HTML parser |
| 21:16 | <othermaciej> | at least, I would be very very surprised if this was not true |
| 21:18 | <weinig> | othermaciej: are you claiming that going from UTF-8 to UTF-16 to UTF-8 and back to UTF-16 is slow ;) |
| 21:18 | <othermaciej> | weinig: that's kind of what I am assuming |
| 21:19 | <othermaciej> | (and I hope we can fix it) |
| 21:19 | <hsivonen> | weinig: why do you go once to UTF-8 in between? |
| 21:20 | <weinig> | hsivonen: libxml only operates on UTF-8 |
| 21:20 | <hsivonen> | weinig: oh so does your stream decoder always expand to UTF-16 first? |
| 21:21 | <weinig> | hsivonen: yes |
| 21:21 | <hsivonen> | I see. |
| 23:40 | <dglazkov> | paging Dr. Hixie |
| 23:41 | <heycam> | hsivonen, SVG Tiny 1.2 does say, in the Reference Restrictions section: http://www.w3.org/TR/SVGMobile12/linking.html#ReferenceRestrictions |
| 23:42 | <heycam> | 1.1 doesn't though |
| 23:44 | <heycam> | correction, 1.1 does: http://www.w3.org/TR/SVG11/struct.html#uriReferenceDefinition |